Docjar: A Java Source and Docuemnt Enginecom.*    java.*    javax.*    org.*    all    new    plug-in

Quick Search    Search Deep

AI.NeuralNetworks
Class FeedForwardWeightedTrainingLayer  view FeedForwardWeightedTrainingLayer download FeedForwardWeightedTrainingLayer.java

java.lang.Object
  extended byAI.NeuralNetworks.FeedForwardLayer
      extended byAI.NeuralNetworks.FeedForwardWeightedLayer
          extended byAI.NeuralNetworks.FeedForwardWeightedTrainingLayer
Direct Known Subclasses:
DistributedFeedForwardWeightedTrainingLayer

public class FeedForwardWeightedTrainingLayer
extends FeedForwardWeightedLayer

This class is used to represent the inner and output layers of a multilayer feed forward neural network where the back error propabation algorithm is run


Field Summary
(package private)  float[][] deltaWeights
           
(package private)  float[] error
           
(package private)  float[] outputError
           
 
Fields inherited from class AI.NeuralNetworks.FeedForwardWeightedLayer
weights
 
Fields inherited from class AI.NeuralNetworks.FeedForwardLayer
activation
 
Constructor Summary
FeedForwardWeightedTrainingLayer(FeedForwardWeightedLayer layer)
          Creates a new FeedForwardWeightedTrainingLayer instance form a FeedForwardWeightedLayer instance
 
Method Summary
 void addDeltaWeights(float learningRate)
          Adds the deltaWeights to the weights
 void adjustDeltaWeights(FeedForwardLayer previousLayer, FeedForwardWeightedTrainingLayer nextLayer)
          Calculates and adds the delta weights for this pattern
 void adjustDeltaWeights(float[] expectedOutput, FeedForwardLayer previousLayer)
          Calculates and adds the delta weights for this pattern
 float[] getError()
          returns the error of the layer
 float[] getOutputError()
           
 void resetDeltaWeights()
          Sets the acumulated delta weights to 0
 void resetError()
          Sets the error of all this layer to 0
 
Methods inherited from class AI.NeuralNetworks.FeedForwardWeightedLayer
getWeights, runLayer, sigmoid, toString
 
Methods inherited from class AI.NeuralNetworks.FeedForwardLayer
getActivation, size
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait
 

Field Detail

deltaWeights

float[][] deltaWeights

error

float[] error

outputError

float[] outputError
Constructor Detail

FeedForwardWeightedTrainingLayer

public FeedForwardWeightedTrainingLayer(FeedForwardWeightedLayer layer)
Creates a new FeedForwardWeightedTrainingLayer instance form a FeedForwardWeightedLayer instance

Method Detail

resetDeltaWeights

public void resetDeltaWeights()
Sets the acumulated delta weights to 0


adjustDeltaWeights

public void adjustDeltaWeights(float[] expectedOutput,
                               FeedForwardLayer previousLayer)
Calculates and adds the delta weights for this pattern


adjustDeltaWeights

public void adjustDeltaWeights(FeedForwardLayer previousLayer,
                               FeedForwardWeightedTrainingLayer nextLayer)
Calculates and adds the delta weights for this pattern


resetError

public void resetError()
Sets the error of all this layer to 0


getError

public float[] getError()
returns the error of the layer


getOutputError

public float[] getOutputError()

addDeltaWeights

public void addDeltaWeights(float learningRate)
Adds the deltaWeights to the weights